• Keine Ergebnisse gefunden

The Spatio-Temporal Distribution of Fog in French Guiana as a Precondition for the Appearance of a new Vegetation Type - the Tropical Lowland Cloud Forest

N/A
N/A
Protected

Academic year: 2021

Aktie "The Spatio-Temporal Distribution of Fog in French Guiana as a Precondition for the Appearance of a new Vegetation Type - the Tropical Lowland Cloud Forest"

Copied!
153
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

The Spatio-Temporal Distribution of Fog

in French Guiana as a Precondition for

the Appearance of a new Vegetation Type

- the Tropical Lowland Cloud Forest

kumulative Dissertation zur Erlangung des Doktorgrades

der Naturwissenschaften (Dr. rer. nat.)

dem Fachbereich Geographie der Philipps-Universität Marburg

vorgelegt von

André Obregón Flores aus Düsseldorf

(2)

Vom Fachbereich Geographie

der Philipps-Universität Marburg als Dissertation am 13. Januar 2016 angenommen.

Erstgutachter: Prof. Dr. Jörg Bendix Zweitgutachter: Prof. Dr. Georg Miehe

(3)

Morning Fog in Les Nouragues Nature Reserve, French Guiana Cover Photo of Ecotropica 18 (2012)

(4)
(5)
(6)
(7)

Preface

The fog is finally lifting after years of devoting parts of my life to this work. Finalization of this thesis could not have been achieved without the utmost support of my supervisor Jörg Bendix to whom I am very grateful for his counsel and confidence in me throughout all these years. He and Rob Gradstein have made this project possible and I thank them for giving me the opportunity of becoming a part of it. Funding by the German Research Foundation under grants BE 1780/13-1 and GR 1588/12-1 is gratefully acknowledged. I am also much obliged to Jan Cermak for giving me guidance and motivation in an early stage of this work.

This endeavour would not have been possible without my co-workers in French Guiana. Rütger Rollenbeck provided tremendous help with setting up the climate station and I benefitted substantially from his scientific and technical expertise. Sebastian Achilles is thanked for his enthusiasm and tireless support during several field campaigns under challenging working conditions. I thank Christine Gehrig for providing her botanical expertise to this work and her help in the field, together with Felix Normann and Patrick Weigelt, who have been excellent company during our field campaign in Saül. I am very grateful to Philippe Gaucher (CNRS Guyane) for logistic help in French Guiana and maintenance of the COPAS climate station. I also express my gratitude to Alex Pardow and Michael Lakatos for our fruitful cooperation and their assistance in the field.

The people of French Guiana are thanked for their friendliness and help they offered during this time. I am grateful for having been given the chance to live and work in a nearly pristine tropical rain forest ecosystem for several months which enabled me to appreciate even more the true beauty of nature.

I thank my former colleagues at the Laboratory for Climatology and Remote Sensing for the beautiful working environment and all the time that we spent together in and outside the office. With special gratitude I thank Helga Nitsche, Peer Hechler, Andrea Kaiser-Weiss and Frank Kaspar from the German Meteorological Service (DWD) for facilitating my pursuit of this thesis during my time at DWD. Also, I wish to thank my current Director Barbara Ryan and my

(8)

colleagues at the Group on Earth Observations (GEO) Secretariat in Geneva. I thank Gary Geller for his constructive advice.

Special thanks go to my friends and family for their backing and support whenever needed. I thank my mother for her never ending faith in my capability to reach my objectives.

Finally, I thank my beloved wife Karol. Without her unconditional encouragement and support this thesis could not have been written. I also wish to express my heartfelt love to my son Lionel to whom I dedicate this work.

(9)

Contents

List of Figures ... III List of Tables ... IX List of Acronyms and Symbols ... XI

1 Introduction ... 1

1.1 Fog in tropical lowland forests 1 1.2 Study area 4 1.3 Aims of this work and outline 6 2 Conceptual Design ... 13

3 Canopy level fog occurrence in a tropical lowland forest of French Guiana as a prerequisite for high epiphyte diversity ... 19

3.1 Introduction 21 3.2 Materials and methods 22 3.2.1 Study area ... 22

3.2.2 Experimental setup ... 24

3.2.3 Data analysis ... 25

3.3 Results 27 3.3.1 Fog frequency and fog density ... 27

3.3.2 Relevant meteorological parameters of fog persistence ... 30

3.3.3 Radiative cooling, katabatic flows and fog occurrence ... 32

3.3.4 Fog dynamics ... 34

3.3.5 Fog occurrence and species richness at LCF and LRF sites .. 36

3.4 Discussion and conclusions 39 4 Epiphyte Biomass and Canopy Microclimate in the Tropical Lowland Cloud Forest of French Guiana ... 51

4.1 Introduction 53 4.2 Methods 54 4.2.1 Study area ... 54

4.2.2 Sampling of biomass ... 54

4.2.3 Estimation of epiphyte cover ... 55

4.2.4 Microclimate and fog measurements ... 55

(10)

4.3 Results 56

4.3.1 Epiphytic biomass ... 56

4.3.2 Epiphyte cover ... 57

4.3.3 RH and temperature... 58

4.3.4 Relation of RH and fog events ... 60

4.4 Discussion 61 5 Diversity and vertical distribution of filmy ferns as a tool for identifying the novel forest type “tropical lowland cloud forest” ... 67 5.1 Introduction 69 5.2 Methods 70 5.2.1 Study area ... 70 5.2.2 Epiphyte sampling ... 71 5.2.3 Microclimate measurements ... 73 5.2.4 Statistical analysis ... 73 5.3 Results 73 5.3.1 Species richness ... 73 5.3.2 Species composition ... 75 5.3.3 Vertical distribution ... 75 5.3.4 Canopy microclimate ... 76 5.4 Discussion 79 6 The potential distribution of tropical lowland cloud forest as revealed by a novel MODIS-based fog/low stratus night-time detection scheme ... 87

6.1 Introduction 89 6.2 Study area 91 6.3 Data and methods 94 6.3.1 Fog measurements for training and validation ... 95

6.3.2 Satellite data and preprocessing ... 95

6.3.3 Extraction of topographic features... 96

6.3.4 Radiative transfer calculations ... 98

6.3.5 Discriminant analysis ... 98

6.3.6 Fog detection scheme ... 99

6.3.7 Validation approach ... 101

6.4 Results 102 6.4.1 Suitability of thresholds ... 102

6.4.2 Validation of the novel FLS classification scheme ... 105

6.4.3 Spatial fog occurrence in French Guiana ... 107

6.5 Discussion and Conclusions 109 7 Summary and Outlook ... 124

7.1 Summary 124 7.2 Outlook 127 Zusammenfassung ... 130

(11)

List of Figures

Figure 1.1. Map of the study area. The dotted rectangle marks the inner study area, which comprises the Saül and COPAS sites ... 6 Figure 1.2. Outline of this work (indicating chapters and hypotheses) ... 8 Figure 2.1. Experimental setup for the work packages (WP1: COPAS

Climate Station, WP2: Mobile measurements, WP3: Remote sensing data) ... 14 Figure 2.2. Overview on instruments, data and methods used for Work

Packages 1-3 ... 17 Figure 3.1. Measurement approach at the study site. (a) Location of the

study site; (b) layout and visualization of hypothesis; (c) placement of mobile sensors in the canopy; (d) permanent measurements on COPAS platform ... 23 Figure 3.2. General streamflow patterns in 10m (sigma-995 level) from

NCEP reanalysis data at different times of day for a) dry season (August–October 2007) and b) rainy season (May–July 2008) ... 24 Figure 3.3. (a) Diurnal course of fog frequency in the dry (mid-August

until mid-November 2007) and rainy (mid-June until mid-July 2008) season at the study site (present weather sensor) (b) diurnal course of average relative air humidity (RH) and specific saturation deficit (sd) for the rainy and dry season, both at the canopy–atmosphere interface level ... 28 Figure 3.4. Frequency of visibility classes for the fog at the study site

(present weather sensor) at canopy–atmosphere interface level ... 29 Figure 3.5. Box–Whisker plots of fog persistence shown for three fog

density classes (dense fog: visibility <200 m, moderate fog: visibility between 200 and 500 m, light fog: visibility between 500 and 1000 m) in the dry season. Fog density (mean visibility) is calculated for each corresponding fog period ... 29 Figure 3.6. Median values of global solar radiation (RG) under foggy and

(12)

the morning period from 6 to 9 LST (dotted line: 1:1 bisection); black line: regression line (r2 = 0.822) for dense fog (visibility <200 m); grey line: regression line (r2 = 0.937) for light and moderate fog (visibility 200–1000 m) ... 30 Figure 3.7. Average diurnal course of air temperature in two tree levels

(canopy, innercrown) and wind direction (COPAS tower) on hill and in valley at the study site (field campaign March–April 2008) ... 32 Figure 3.8.. a) Distribution of wind directions (COPAS tower) in dry

season for days of long (>6 h, high fog) and short fog persistence (<2 h, low fog) in the evening hours (19:00–00:00 LST) and b) terrain map of the surroundings ... 34 Figure 3.9. Fog dynamics for the period of 20–24 October 2007 ... 35 Figure 3.10. Mean diurnal course of relative humidity in two tree levels

(canopy, inner crown) and leaf wetness duration (percentage of time per hour) on hill and in valley at the study site (field campaign March–April 2008) ... 37 Figure 3.11. Component plot for the principal component analysis using air

temperature, relative humidity and epiphyte diversity parameters ... 37 Figure 4.1. Abundance of all epiphytes in percentage of bark coverage in

lowland cloud forest (LCF) (white) and lowland rain forest (LRF) (gray) in different height zones (I–VI) on the tree; N= 24 trees per forest type. Boxes indicate upper and lower quartile of data, unbroken line gives the median, dotted line the mean and whiskers 5th/95th percentile. Levels of significance are obtained with unpaired t-tests and shown by asterisks, *P<0.05, **P<0.01, ***P<0.001 ... 57 Figure 4.2. Abundance of epiphytic lichens, bryophytes, and angiosperms

in percentage of bark coverage on trunks (zones I–III) and crowns (zones IV–VI) in lowland cloud forest (LCF) (white) and lowland rain forest (LRF) (gray); N= 72 estimates per forest type. Boxes indicate upper and lower quartile of data, unbroken line gives the median, dotted line the mean and whiskers 5th/95th percentile. Levels of significance are obtained with unpaired t-tests and shown by asterisks, *P<0.05, **P<0.01, ***P<0.001 ... 58 Figure 4.3. Diurnal course of vapor pressure deficit (VPD), calculated

using hourly mean temperature and air humidity during 48 d of the dry season in lowland cloud forest (LCF) (white circles) and lowland rain

(13)

forest (LRF) (gray circles). Levels of significance are obtained with unpaired t-tests and shown by asterisks, (*P<0.05, **P<0.01, ***P<0.001 .... 59 Figure 4.4. Relative (in percentage) and absolute (in N= number of total

events) occurrence of light fog (visibility<1 km, light gray) and dense fog (visibility<0.2 km, dark gray) for classes of high relative humidity (RH) in lowland cloud forest at Canopy Operating Permanent Access System station, June–July 2008 ... 60 Figure 5.1. Number of epiphytic Hymenophyllaceae species per height zone

in lowland cloud forest and lowland rain forest; n = 16 trees per forest type. Asterisks indicate level of probability calculated with unpaired t-test (* p<0.05, ** p<0.01). z1 = trunk base, z2 = lower trunk, z3 = upper trunk, z4 = lower canopy, z5 = middle canopy, z6 = outer canopy ... 75 Figure 5.2. Correlation between number of Hymenophyllaceae species per

tree and mean diurnal relative air humidity for 60 days of the dry season in the canopy of 7 trees in lowland cloud forest. Unbroken line indicates the regression, dotted line the 0.95 confidence interval. R = 0.82, P<0.05 ... 76 Figure 5.3. Evaporation per day in the inner and outer canopy of lowland

cloud forest and lowland rain forest, calculated using a simplified version of the Penman equation provided by Valiantzas (2006), incorporating daily global radiation, relative humidity, air temperature and the latitude of the site. Boxes indicate upper and lower quartile of data, unbroken line indicates the median, dotted line the mean, whiskers 5th/95th percentile, and circles mark outliers ... 77 Figure 5.4. Mean diurnal course of global radiation and relative humidity

in lowland cloud forest and lowland rain forest in the inner and outer canopy during 22 September – 24 October 2007 ... 78 Figure 5.5. Vertical distribution of two Hymenophyllaceae species in

lowland rain forest and lowland cloud forest, expressed by percent occurrence per zone as compared with occurrence on whole trees. z2 = lower trunk, z3 = upper trunk, z4 = lower canopy, z5 = middle canopy, z6 = outer canopy ... 80 Figure 6.1. Terrain map of French Guiana and the inner study area

(dotted rectangle, refer to figure 6.3 for details). Digital Elevation Model derived from the Shuttle Radar Topographic Mission (USGS, 2006).The marked dots represent two study sites (COPAS, Saül) intensively investigated regarding climate and epiphytic diversity ... 92

(14)

Figure 6.2. Flowchart of methodology showing methods with the corresponding sections (grey) and input data (white). Scatt = Scatterometer data, PFI = Potential Fog Index, ΔT = temperature difference IR-MIR band). ... 94 Figure 6.3. Digital Elevation Model for the inner study area (A), slope

position classes derived by the Topographic Position Index for the COPAS area at 100 m resolution (B), and the Potential Fog Index at 1 km resolution (C). ... 97 Figure 6.4. Proposed fog/low stratus detection scheme. ... 99 Figure 6.5. Confusion matrix and definition of skill scores used for the

validation. ... 101 Figure 6.6. Radiative transfer calculations for a ground-touching fog layer

with different optical thicknesses. Brightness temperature differences (ΔT11-3.9) are shown for three standard atmospheres

(MLW=midlatitudes winter, MLS=midlatitudes summer,

TROP=tropics). The simulated fog layer in the study area corresponds to an optical thickness of around 1. ... 102 Figure 6.7. (Left): RTC-derived functions of brightness temperature

differences vs. total precipitable water for different cloud fractions. Points indicate the values derived by the radiative transfer calculations. An exponential fit was used for the regression lines. The gray line shows the empirical relationship between ∆T11-3.9 and TPW using CALIOP clear vs. cloudy determinations and MODIS radiance data (taken from Ackermann et al., 2010). The function is calculated as ΔT =-0.0077+1.1234*TPW+(-0.3403*TPW^2). The vertical line marks the TPW value (=4 cm) of which above modified PFI functions are used. (Right): Threshold functions for the separation of fog and fog-free pixels (according to the in-situ fog measurements) derived by radiative transfer calculations (table 6.4, PFI=0.8) and discriminant analysis (discriminant line: ΔT =-1.77TPW+6.1). ΔT11-3.9 values of MODIS data of all days with validation data are shown for the corresponding MODIS-derived TPW values. ... 104 Figure 6.8. ROC diagrams showing validation results for the COPAS pixel

(PFI=0.79) for the dry season 2007 (N=29) (top) and all available validation days (N=80) (bottom). The dashed lines indicate retrievals with variable thresholds depending on TPW using different RTC-derived PFI functions (PFI in %). The solid lines indicate fixed thresholds

(15)

between -1.5 and +1. Since the fog detection scheme is a discrete classifier, POD and FAR correspond to a single point in ROC space. ... 106 Figure 6.9. Fog frequency maps for the three different retrievals in the

COPAS area: A) fixed threshold of -0.52; B) variable TPW and fixed PFI; C) variable TPW and variable PFI. The dry season 2007 is presented, which has the best coverage of validation data. Hillshading is enabled using SRTM data. ... 107 Figure 6.10. Fog frequency map for August, September, October (dry

season) from 2007 till 2010 encompassing a detailed inset of the inner area covering the sites of the field measurements. A dynamical threshold based on variable TPW and variable PFI was used. SRTM data was applied for hillshading. ... 108

(16)
(17)

List of Tables

Table 3.1. Pearson´s correlation coefficients of fog persistence (VIS ≤ 1 km) at canopy level of the COPAS station in relation to meteorological parameters for time frames before (8:00-19:00 LST) and during fog formation (19:00-0:00 LST). Correlations are not significant if no p-value is given; α = 5%. Observation period Aug-Nov 2007 (dry season, 98 days) and Jun-Jul 2008 (rainy season, 34 days) ... 31 Table 3.2. Site comparison (median and median absolute deviation) of

meteorological parameters measured during field campaign in March-April 2008 during nighttime (0-8 LST) and botanical data collected in 2007 (8 trees per site). No visibility data (*) was available for the hill site ... 36 Table 3.3. Principal component analysis using air temperature, relative

humidity and diversity measured in 16 trees (eight in valley, eight in hill forest). Total variance explained and component matrix for eight parameters... 38 Table 4.1 Comparison of microclimatological factors measured in the

middle canopy of lowland cloud forest and lowland rain forest of central French Guiana. RH= daily relative air humidity, t= daily air temperature. Top: 48 days of dry season September-October 2007. Bottom: 230 days of wet season November 2007-mid June 2008 ... 59 Table 5.1. Tree height and diameter at breast height (dbh) of trees

sampled in lowland cloud forest (LCF) and lowland rain forest (LRF) in central French Guiana. Tree species name and family are provided where available ... 72 Table 5.2. Occurrence and vertical distribution of Hymenophyllaceae in

lowland cloud forest (LCF) and lowland rain forest (LRF) in central French Guiana. Numbers refer to the number of samples in which the species was recorded. For further explanation see text. z1 = trunk base, z2 = lower trunk, z3 = upper trunk, z4 = lower canopy, z5 = middle canopy, z6 = outer canopy. Nomenclature of taxa follows Lellinger (1994) and Cremers (1997) ... 74

(18)

Table 6.1. Summary of field investigations regarding performance of canopy epiphytes in lowland cloud forest (LCF) and lowland rain forest (LRF) for the two sites marked in figure 6.1. The data are compiled from Gehrig-Downie et al. (2011, 2012 & 2013). ... 90 Table 6.2. Classification of topographic position index (TPI) values into

slope position classes using thresholds derived by the standard deviation (SD) of TPI. ... 97 Table 6.3. Results for brightness temperature differences (∆T11-3.9 µm) from

radiative transfer calculations for a constant effective radius of 4 μm (= fog) and a geometrical thickness of 30 m for different potential fog coverages (expressed as PFI) and three standard atmospheres

(MLW=midlatitudes winter, MLS=midlatitudes summer,

TROP=tropics)... 103 Table 6.4. Functions of ΔT and TPW derived by radiative transfer

calculations for different cloud fractions using an exponential fit. Corresponding PFI classes are shown. ... 104

(19)

List of Acronyms and Symbols

AVHRR Advanced Very High Resolution Radiometer

CALIOP Cloud-Aerosol Lidar with Orthogonal Polarization

COPAS Canopy Operating Permanent Access System

dbh diameter at breast height [cm]

DEM Digital Elevation Model

ERA ECMWF Reanalysis

FAO Food and Agriculture Organization

FAR False Alarm Ratio

FG French Guiana

FLS Fog / Low Stratus

GOES Geostationary Operational Environmental Satellite

IR Infrared

ITCZ Intertropical Convergence Zone

LBA Large-Scale Biosphere-Atmosphere Experiment in

Amazonia

LCF Tropical Lowland Cloud Forest

Lidar Light detection and ranging

LRF Tropical Lowland Rain Forest

LST Local Solar Time

LUT Lookup Table

LWC Liquid Water Content [g m−3]

LWD Leaf Wetness Duration [h]

MCF Tropical Montane Cloud Forest

MIR Mid Infrared

MODIS Moderate Resolution Imaging Spectroradiometer

NCEP-NCAR National Centers for Environmental Prediction / National Center for Atmospheric Research

NOAA National Oceanic and Atmospheric Administration

PCA Principal Component Analysis

PFI Potential Fog Index

(20)

Radar Radio detecting and ranging

RH Relative Humidity [%]

ROC Receiver Operating Characteristics

RTC Radiative Transfer Calculations

sd Specific saturation deficit [g kg-1]

SRTM Shuttle Radar Topographic Mission

SZA Satellite Zenith Angle

TMCF Tropical Montane Cloud Forest

TPI Topographic Position Index

TPW Total Precipitable Water

USGS United States Geological Survey

UTM Universal Transverse Mercator

VIS Horizontal Visibility [m]

VPD Vapour Pressure Deficit [hPa]

WMO World Meteorological Organization

(21)

1

Introduction

1.1

Fog in tropical lowland forests

Cloud forests are a type of rainforest which is under the permanent or regular influence of clouds. They are mainly associated with tropical mountain regions, being well-known as Tropical Montane Cloud Forests (TMCFs) (Bruijnzeel et al. 2010), but they are also found in subtropical and temperate regions (Weathers 1999).

When a cloud is in direct physical contact with the underlying surface, it is meteorologically referred to as fog. Several definitions of fog are found in the literature, either defining fog by thresholds of horizontal visibility or as a physical phenomenon (Cermak 2006). The World Meteorological Organization defines fog as a suspension of very small, usually microscopic water droplets in the air reducing the horizontal visibility at the Earth´s surface to less than 1 km (WMO 1992). In synoptic reporting, fog is generally treated as an obstruction to visibility. Most international studies follow the definition of fog by horizontal visibility. Glickman (2000) extended this definition by introducing the condition of small cloud droplets (diameter below 200 μm) being present.

Physically, a fog consists of very small water droplets with a diameter ranging from 1 to 40 μm, with fall velocities below 5 cm s-1 (WMO 2010). Fog can form via different meteorological processes, involving either cooling of surface air or evaporation from a warm and moist surface, or a combination of both. Air humidity close to saturation, low wind velocities and a reduction of turbulence are important prerequisites of fog formation (Findlater 1985, Schilling 1991, Pasricha et al. 2003). Related to the meteorological processes involved in the formation of fog, a broad variety of definition of fog types exist. Bruijnzeel et al. (2005) proposed seven types of fog, which are named according to the meteorological process (radiation fog, sea fog, steam fog, advection fog) or to the geographic location (coastal fog, valley fog, mountain fog). However, a sharp separation remains difficult, e.g. valley fog is typically a radiation fog and coastal fog an advection fog.

Radiation fog is one of the most common fog types related to cooling of air temperature below the saturation point. It is predominantly formed in valleys

(22)

1 Introduction

and basins due to nocturnal outgoing radiation and cooling of the boundary layer air, often promoted by katabatic flows. Advection fog occurs when humid air is transported over a colder surface leading to cooling of the air mass and condensation (Roach 1995). Cooling of an air mass can also occur by forced lifting along mountain slopes until reaching a condensation level which leads to orographic fog, also known as mountain fog or cloud fog. Since this fog type refers to all kind of fogs at high elevation it is not necessarily a stratus or stratiform cloud (Eugster 2008).

Fog as a research object is extensively studied by different research communities comprising a multiplicity of fields, ranging from meteorological studies (e.g. nowcasting), socio-economical applications (e.g. water harvesting) and ecological studies (e.g. biodiversity). Reviews on past achievements and future perspectives of fog research are found in Gultepe et al. (2007) and Eugster (2008). While fog is often associated with negative attributes (e.g. for traffic and human well-being) it has remarkable positive interrelations with the biotic and abiotic environment (Bendix et al. 2011). Although meteorologists are generally more concerned with fog as an obstruction to visibility, it is important to note that fog water deposition to surfaces is a form of precipitation (WMO 2008) which can be of great ecological value.

From a hydrological perspective, cloud moisture input can make a considerable contribution to the water balance of tropical watersheds (Stadtmüller & Agudelo 1990). The crucial role of fog as an additional water supply source, in particular for epiphytic plants, is widely accepted (Bruijnzeel et al. 2011). The process of fog water deposition is also known as fog drip, horizontal precipitation, cloud water interception, cloud stripping, lateral cloud filtering or occult precipitation. Regarding the role of fog in plant ecology it should be stressed that tropical montane cloud forest trees are capable to absorb cloud water directly through their leaves - a process which is described as foliar uptake (Goldsmith et al. 2013).

In addition to canopy moistening, other effects of cloud immersion that are beneficial for epiphytic vegetation that rely on moist environments and water supply from the air include the reduction of solar radiation, vapour deficit and evapotranspiration (Hamilton 1995). Hence, fog occurrence is regarded as the single most important microclimatic feature affecting the distribution and function of TMCF plants (Oliveira et al. 2014), and particularly as a major precondition for high epiphyte diversity (e.g. Bruijnzeel et al. 2011). Richards et al. (1996) and many others thereafter showed that TMCF may harbour more species of epiphytes than any other forest type.

Since meteorological conditions creating a foggy environment are not restricted to mountainous areas, cloud forest can occur at virtually any elevation. In fact, their occurrence is also documented for lowlands in temperate regions,

(23)

1.1 Fog in tropical lowland forests e.g. at the coast of California (Dawson 1998) and in northern Chile (Aravena et al. 1989). However, lowland cloud forests in the tropics have remained largely under-investigated.

It is a general assumption that montane forests are richer in species than lowland forests (e.g. Richards 1984, Hamilton & Bruijnzeel 2000). However, botanic inventories of non-vascular epiphyte diversity (bryophytes, lichens) in tropical lowland and montane forests across the Neotropics give reason to call this into question. For example, in a moist lowland forest in Guyana, five trees yielded 88 moss species (50 liverworts) and in Saül (French Guiana), 40 species of moss and 60 species of liverwort were found on only four trees. One single tree yielded up to 50 bryophyte species (Cornelissen & Gradstein 1990, Montfoort & Ek 1990, Gradstein 1995, 2006). These figures are the highest worldwide reported for tropical lowland rain forest. Species richness of epiphytic liverworts in French Guiana has reported to be three times higher than in Amazonian lowland forest at Surumoni (Venezuela) (Gradstein 2006) and up to 1.5 times higher than in moist submontane and montane forest in the tropical Andes (Wolf 1993, Acebey et al. 2003).

The causes of high species richness in Guyana and French Guiana have been unclear because elevation and annual precipitation are similar for instance to the Surumoni site which has lower epiphyte richness and abundance (Gradstein 1995, Anhuf & Winkler 1999). Gradstein (2006) pointed at the high incidence of fog in French Guiana which is absent at Surumoni as a potential reason. The high species richness was hence explained by the favourable air moisture regime caused by fog which allows poikilohydric species to thrive in the hot lowland forest. Fog during day times may prevent desiccation and allow these plants to achieve positive net photosynthesis in spite of high temperatures (Gradstein 2006).

To date, no comprehensive studies on either the spatio-temporal occurrence of tropical canopy fog or on the meteorological processes leading to fog formation have been conducted in tropical lowland forest areas. The role of fog in the distribution of tropical lowland forest types remains poorly understood. Investigations are made more difficult by the fact that measurements of horizontal visibility at the surface/above the canopy in tropical forest are rarely available. Therefore, it is unknown which fog type might foster the epiphyte-rich lowland forest in valleys of central French Guiana.

Prior to describing the objectives of the present study and the conceptual approach of this work in subsequent sections, the next section gives a brief overview of the study area.

(24)

1 Introduction

1.2 Study area

French Guiana is located on the northeastern coast of South America. It lays upon the easternmost part of the Guiana Shield, a cratone of the South American plate, which comprises the whole of French Guiana, Suriname and Guyana as well as parts of Venezuela, Brazil and Colombia. The Guiana Shield extends roughly between the Orinoco River in the west and the Amazon River in the south. It is one of the least populated areas of the world and large areas are covered with almost undisturbed rain forests (Hammond 2005a).

Hammond (2005b) distinguished five large landforms for the Guiana Shield of which the “Precambrian Rolling Hills” account for almost all of French Guiana. This landform describes an undulating granitoid landscape that has been created through synclinal folding and differential weathering (Hammond 2005b). The terrain in French Guiana is undulating throughout the country; in particular, the central part is hilly and considerably dissected. The elevation of hills, ridges and small river valleys varies between 50 and 300 m a.s.l. (refer to figure. 1.1).

According to the Global Forest Resources Assessment 2010 (FAO 2010), French Guiana has a total forest area of around 8 million hectares and its relative forest cover ranks among the highest worldwide. 98% of the land area is covered by forests, of which 95% are classified as old-growth or primary forests. Between 1990 and 2010 total forest extent decreased by 1.3 % and primary forest decreased by 3.9%. Above-ground biomass and carbon stock per land area in French Guiana are among the highest worldwide (Saatchi 2011).

French Guiana has a tropical climate, with annual mean daily temperatures between 25° and 27°C. Annual rainfall ranges from around 2,500 mm in the west to 3,500 mm at the Atlantic Coast with a bimodal seasonal precipitation cycle. The main dry season is between August and November. A less pronounced dry period occurs in February and March. Bovolo et al. (2011) give an overview of the fine-scale regional climate in the Guianas using ERA-40 reanalysis data for 1958-2001.

In terms of meteorological measurements, the entire Guiana Shield is one of the poorest monitored regions in the world (Hammond 2005b). In French Guiana, Météo France operates six meteorological stations of which five are located close to the Atlantic coast. The station Maripasoula is located at the border to Suriname, 50 km up the Maroni River. There are no official inland stations in French Guiana.

However, long-term meteorological measurements for research purposes have been conducted in the Nouragues Natural Reserve, around 100 km inland from

(25)

1.2 Study area the Atlantic coast. Grimaldi & Riera (2001) give a summary of these measurements for November 1987 to December 1996 (refer also to chapter 3.2.1 of the present study).

Figure 1.1 gives a topographic overview of the study area. The two sites where field work has been conducted are briefly described in the following:

1) COPAS site:

The Canopy Operation Permanent Access System (COPAS) (refer to next chapter) is located close to the research station Saut Pararé, within the uninhabited Nouragues Natural Reserve (4° 2’ 30’’ N, 52° 40’ 30’’W, 75 m a.s.l.). It is situated in the valley of the Arataye river, 350 m away from the edge of the river. The site is ca. 100 km inland from Atlantic coast. Elevation varies from 50 to 200 m a.s.l.. The area comprises high forest, slope forest and forest growing on hydromorphic soil. Canopy trees reach up to approximately 50 m.

2) Saül site:

The secondary site is located in the vicinity of the village of Saül (3°37'20"N, 53°12'31"W), about 200 km southwest of the Atlantic coast at the headwaters of three of French Guiana’s major river systems. The area is hilly and considerably dissected, with small river valleys at about 200–250 m and hills to about 400m in elevation. The area is covered by mixed lowland rain forest (e.g. De Granville 1986, 2001, Mori & Boom 1987). The forest canopy varies in height from 20 to 45 m, with emergent trees reaching up to 55 m (Mori & Boom 1987).

(26)

1 Introduction

Figure 1.1. Map of the study area. The dotted rectangle marks the inner study area, which comprises the Saül and COPAS sites

The study sites have been selected for their representativeness (each site harbouring nearly undisturbed valley and hill forests) and due to their sufficient distance to the coast (to avoid confusion with coastal fog). For detailed descriptions of the study sites the reader may refer to the individual chapters of this thesis.

1.3

Aims of this work and outline

The present work is part of a collaborative study between the University of Göttingen and the University of Marburg. The primary goal of this collaboration is to investigate the relationship between climate, especially fog, and epiphyte vegetation in French Guiana. The underlying presumption of the study is the existence of a novel type of tropical lowland cloud forest (LCF), which is caused by frequent fog occurrence and characterized by high epiphyte diversity.

Hitherto, no systematic investigation has been available on the interaction of energy balance, local breeze pattern and fog occurrence in tropical lowland forests. The knowledge of temporal and spatial fog dynamics in the tropical lowland is generally poor. The main goal of this thesis is therefore to provide an in-depth study of the fog phenomenon and its spatiotemporal dynamics in French Guiana and to assess its implications on the formation of cloud forest in the tropical lowland.

(27)

1.3 Aims of this work and outline Two main hypotheses will be tested within this work:

H1 Radiation fog occurs frequently in valleys of French Guiana and is supported by nocturnal katabatic flows. Valley fog is a regularly occurring phenomenon in space.

H2 Valley forest is characterized by high epiphyte diversity, abundance and biomass due to a favourable canopy microclimate. The occurrence of fog thus spatially determines the new vegetation unit Tropical Lowland Cloud Forest.

Following from these hypotheses, the aims of the presented work are:

 To analyse the underlying meteorological processes leading to fog formation

 To investigate microclimatic differences between forest in valleys and ridges

 To shed light on the relation between fog occurrence and epiphyte species richness in LCF

 To reveal the spatio-temporal dynamics of night-time fog throughout French Guiana and, with this, to provide evidence for potential LCF habitats

The investigation should shed light on current knowledge deficits:

 Fog formation has never been studied in depth in tropical lowland forests and therefore its spatial occurrence and role in these habitats is poorly understood

 The spatio-temporal distribution of fog can only be detected with

weather satellite data. However, fog in French Guiana is hypothesized to be a small-scale phenomenon, frequently restricted to narrow river valleys which often lie in the sub-pixel detection range even of polar-orbiting weather satellites with relatively high spatial resolution (mostly 1 km). A subpixel fog/low stratus detection scheme is necessary which can only rely on IR satellite data for night/early morning time applications.

Figure 1.2 presents an outline of this work. It starts in chapter 2 with the description of the conceptual design including an overview of the work packages and the technical preparation. The implementation of the work packages is embedded in four publications: The underlying meteorological processes of fog

(28)

1 Introduction

formation in French Guiana are presented in the publication Obregon, Gehrig-Downie, Gradstein, Rollenbeck & Bendix 2011: Canopy Level Fog Occurrence in a Tropical Lowland Forest of French Guiana as a Prerequisite for High Epiphyte

Diversity, Agricultural and Forest Meteorology, doi:

10.1016/j.agformet.2010.11.003 (chapter 3.). The investigation of canopy microclimate in different habitats and the impacts on the epiphytic vegetation are embedded in two publications: Gehrig-Downie, Obregon, Bendix, Gradstein 2011, Epiphyte Biomass and Canopy Microclimate in the Tropical Lowland Cloud Forest of French Guiana, Biotropica, doi: 10.1111/j.1744-7429.2010.00745.x (chapter 4) and Gehrig-Downie, Marquardt, Obregon, Bendix, Gradstein 2012, Diversity and Vertical Distribution of Filmy Ferns as a Tool for Identifying the Novel Forest Type “Tropical Lowland Cloud Forest”, Ecotropica, 18, 35-44 (chapter 5). Finally, the development of an algorithm to detect the spatio-temporal fog dynamics in French Guiana is presented in the publication Obregon, Gehrig-Downie, Gradstein & Bendix 2014: The potential distribution of tropical lowland cloud forest as revealed by a novel MODIS-based fog/low stratus night-time detection scheme, Remote Sensing of Environment, doi: 10.1016/j.rse.2014.09.005 (chapter 6). A short summary of the results and an outlook provided in chapter 7 conclude this thesis.

(29)

References

References

Acebey, A., Gradstein, S.R., Kromer, T., 2003. Species richness and habitat diversification of bryophytes in submontane rain forest and fallows of Bolivia. Journal of Tropical Ecology 19, 9-18.

Aravena, R., Suzuki, O., Pollastri, A., 1989. Coastal fog and its relation to groundwater in the IV region of northern Chile. Chemical Geology: Isotope Geoscience section 79, 83-91.

Bendix, J., Eugster, W., Klemm, O., 2011. Fog - boon or bane? Erdkunde 65, 229-232.

Bovolo, C.I., Pereira, R., Parkin, G., Kilsby, C., Wagner, T., 2012. Fine-scale regional climate patterns in the Guianas, tropical South America, based on observations and reanalysis data. International Journal of Climatology 32, 1665-1689.

Bruijnzeel, L.A., Eugster, W., Burkard, R., 2005. Fog as a hydrologic input, in: Anderson, M.G. (Ed.), Encyclopedia of Hydrological Sciences. John Wiley, Chichester, UK, pp. 559-582.

Bruijnzeel, L.A., Kappelle, M., Mulligan, M., Scatena, F.N., 2010. Tropical montane cloud forests: state of knowledge and sustainability perspectives in a changing world, in: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, Cambridge, UK, pp. 691-740.

Bruijnzeel, L.A., Mulligan, M., Scatena, F.N., 2011. Hydrometeorology of tropical montane cloud forests: emerging patterns. Hydrological Processes 25, 465-498. Cermak, J., 2006. SOFOS – A New Satellite-based Operational Fog Observation

Scheme, Philipps-Universitat Marburg, Germany.

Cornelissen, J.H.C., Gradstein, S.R., 1990. On the occurrence of bryophytes and macrolichens in different lowland rain forest types of Mabura Hill, Guyana. Tropical Bryology 3, 29-35.

Dawson, T.E., 1998. Fog in the California redwood forest: Ecosystem inputs and use by plants. Oecologia 117, 476-485.

de Granville, J.J., 1986. Flore et vegetation. Saga, Cayenne, French Guiana. de Granville, J.J., 2001. Végétation, in: Barret, J. (Ed.), Altlas illustré de la

Guyane. IRD Editions, Paris, France, pp. 52-55. Eugster, W., 2008. Fog Research. Erde 139, 1-10.

(30)

References

FAO, 2010. Global Forest Resources Assessment 2000, Rome, Italy.

Findlater, J., 1985. Field investigations of radiation fog formation at outstations. Meteorological Magazine 114, 187-201.

Gehrig-Downie, C., Marquardt, J., Obregon, A., Bendix, J., Gradstein, S.R., 2012. Diversity and vertical distribution of filmy ferns as a tool for identifying the novel forest type “tropical lowland cloud forst”. Ecotropica 18, 35-44. Gehrig-Downie, C., Obregon, A., Bendix, J., Gradstein, S.R., 2011. Epiphyte

Biomass and Canopy Microclimate in the Tropical Lowland Cloud Forest of French Guiana. Biotropica 43, 591-596.

Glickman, T.S., 2000. Glossary of meteorology. 2nd edition. American Meteorological Society, Boston.

Goldsmith, G.R., Matzke, N.J., Dawson, T.E., 2012. The incidence and implications of clouds for cloud forest plant water relations. Ecology Letters, n/a-n/a.

Gradstein, S., 1995. Diversity of Hepaticae and Anthocerotae in montane forests of the tropical Andes, in: Churchill, S.P., Balslev, H., Forero, E., Luteyn, J.L. (Eds.), Biodiversity and Conservation of neotropical Montane Forests. New York Botanical Garden Press, New York, pp. 321-343.

Gradstein, S.R., 2006. The lowland cloud forest of French Guiana – a liverwort hotspot. Cryptogamie Bryologie 27, 141-152.

Grimaldi, M., Riéra, B., 2001. Geography and Climate, in: Bongers, F., Charles-Dominique, P., Forget, P.-M., Théry, T. (Eds.), Nouragues: Dynamics and Plant-Animal Interactions in a Neotropical Rainforest. Kluwer, Dordrecht, The Netherlands, pp. 9-18.

Gultepe, I., Tardif, R., Michaelides, S.C., Cermak, J., Bott, A., Bendix, J., Mueller, M.D., Pagowski, M., Hansen, B., Ellrod, G., Jacobs, W., Toth, G., Cober, S.G., 2007. Fog research: A review of past achievements and future perspectives. Pure and Applied Geophysics 164, 1121-1159.

Hamilton, L.S., 1995. Mountain Cloud Forest Conservation and Research: A Synopsis. Mountain Research and Development 15, 259-266.

Hamilton, L.S., Bruijnzee, L.A., 2000. Decision Time for Cloud Forests. UNESCO Division of Water Sciences, Paris.

Hammond, D.S., 2005a. Ancient land in a modern world, in: Hammond, D.S. (Ed.), Tropical forests of the Guiana shield: ancient forests in a modern world. CABI Publishing, Wallingford, pp. 1-14.

(31)

References Hammond, D.S., 2005b. Biophysical features of the Guiana shield, in: Hammond, D.S. (Ed.), Tropical forests of the Guiana shield: ancient forests in a modern world. CABI Publishing, Wallingford, pp. 15-194.

Montfoort, D., Ek, R., 1990. Vertical distribution and ecology of epiphytic bryophytes and lichens in a lowland rain forest in French Guiana, Institute of Systematic Botany, Utrecht.

Mori, S.A., Boom, B.M., 1987. The forest, in: Mori, S.A. (Ed.), The Lecythidaceae of a Lowland Neotropical Forest: La Fumee Mountain, French Guiana, pp. 9-29.

Obregon, A., Gehrig-Downie, C., Gradstein, S.R., Bendix, J., 2014. The potential distribution of tropical lowland cloud forest as revealed by a novel MODIS-based fog/low stratus night-time detection scheme. Remote Sensing of Environment 155, 312-324.

Obregon, A., Gehrig-Downie, C., Gradstein, S.R., Rollenbeck, R., Bendix, J., 2011. Canopy level fog occurrence in a tropical lowland forest of French Guiana as a prerequisite for high epiphyte diversity. Agricultural and Forest Meteorology 151, 290-300.

Oliveira, R.S., Eller, C.B., Bittencourt, P.R.L., Mulligan, M., 2014. The hydroclimatic and ecophysiological basis of cloud forest distributions under current and projected climates. Annals of Botany 113, 909-920.

Pasricha, P.K., Gera, B.S., Shastri, S., Maini, H.K., John, T., Ghosh, A.B., Tiwari, M.K., Garg, S.C., 2003. Role of the water vapour greenhouse effect in the forecasting of fog occurrence. Boundary-Layer Meteorology 107, 469-482. Richards, P.W., 1984. The ecology of tropical forest bryophytes, in: Schuster,

R.M. (Ed.), New manual of bryology. Hattori Botanical Laboratory, Nichinan, Japan, pp. 1233-1270.

Richards, P.W., Walsh, R.P.D., Baillie, I.C., Greig-Smith, P., 1996. The tropical rain forest an ecological study (2nd edition). Cambridge University Press, Cambridge.

Roach, W.T., 1995. Back to Basics: Fog: Part 2 - The formation and dissipation of land fog. Weather 50, 7-11.

Saatchi, S.S., Harris, N.L., Brown, S., Lefsky, M., Mitchard, E.T.A., Salas, W., Zutta, B.R., Buermann, W., Lewis, S.L., Hagen, S., Petrova, S., White, L., Silman, M., Morel, A., 2011. Benchmark map of forest carbon stocks in tropical regions across three continents. Proceedings of the National Academy of Sciences of the United States of America 108, 9899-9904.

(32)

References

Schilling, V.K., 1991. A Parameterization For Modeling the Meteorological Effects of Tall Forests - A Case-study of A Large Clearing. Boundary-Layer Meteorology 55, 283-304.

Stadtmuller, T., Agudelo, N., 1990. Amount and variability of cloud moisture input in a tropical cloud forest. International Association of Hydrological Sciences Publication 25-32.

Weathers, K.C., 1999. The importance of cloud and fog in the maintenance of ecosystems. Trends in Ecology & Evolution 14, 214-215.

WMO, 1992. International Meteorological Vocabulary, WMO 182. World Meteorological Organization: Geneva, Switzerland, Geneva, Switzerland. WMO, 2008. Guide to Meteorological Instruments and Methods of Observation,

7th ed. World Meteorological Organization, Geneva, Switzerland,.

WMO, 2010. CIMO Guide. World Meteorological Organization, Geneva, Switzerland,.

Wolf, J.H.D., 1993. Diversity Patterns and Biomass of Epiphytic Bryophytes and Lichens Along an Altitudinal Gradient in the Northern Andes. Annals of the Missouri Botanical Garden 80, 928-960.

(33)

2

Conceptual Design

The scientific issues and hypotheses outlined in the previous chapter require a sequence of work packages (WP), which are designed with the following specific aims:

WP1 Analysing the causes of fog formation in tropical lowland forests and the underlying meteorological processes. The aim is to reveal the physical conditions involved in the fog formation, including fog type, fog persistence, fog frequency, and to determine the meteorological parameters triggering the process of fog development.

WP2 Characterizing the microclimatological properties in tropical lowland rain forest (LRF) on slopes and hills, and LCF habitats in valleys and comparing results to the botanical studies. The objective is to confirm the different microclimates of LRF and LCF and to elaborate the impacts for epiphytic vegetation.

WP3 Mapping the spatio-temporal distribution of fog. The purpose is to clarify whether fog formation in tropical lowlands is a local phenomenon or if it is widely distributed in space.

A specific experimental design is needed to address the challenges described above. This specific setup is illustrated in figure 2.1. In the following, the design of each work package including the technical preparation will be described.

Work package 1. The tower system COPAS (Canopy Operation Permanent Access System) at the site Saut Pararé is freely accessible to the international research community (Charles-Dominque et al. 2002) and provides an ideal infrastructure to conduct meteorological measurements with sophisticated instruments directly above the forest canopy at 45 m height. Testing H1 requires the installation of a permanent and automatically operating meteorological station on one platform of the COPAS system to consider the meteorological conditions at canopy level where fog formation is hypothesized to occur.

(34)

2 Conceptual Design

Figure 2.1. Experimental setup for the work packages (WP1: COPAS Climate Station, WP2: Mobile measurements, WP3: Remote sensing data)

The climate station encompasses a set of sophisticated instruments which are designed for accurate measurements under extreme weather conditions. The short-wave and thermal radiation balance is gathered by a net radiometer. A 3D-sonic anemometer is installed to detect low wind velocities at night associated with potential nocturnal drainage flows in order to reveal to what extent katabatic flows play a role in the fog formation process. Further, this type of sensor delivers information about turbulence conditions and stability. Since air humidity is expected to be close to saturation at night, the measurement of humidity is conducted by using a psychrometer. This instrument allows for the accurate calculation of relative humidity close to 100 per cent. The core instrument of the installation is a visibility and present weather sensor. A backscatter receiver enables the calculation of horizontal visibility (and thus fog occurrence) and the detection of precipitation type (fog, haze, drizzle etc.) and liquid water content. Since dewfall may serve as another potential water source

(35)

2 Conceptual Design for epiphytes when fog occurrence or rainfall does not take place, a leaf wetness sensor is installed in addition to the described instruments.

The overall measurement configuration facilitates a full picture of condensation tendencies, fog dynamics and the boundary conditions involved in the fog formation process (wind fields, turbulence). All instruments are operated by solar and battery power and are connected to a data logger computing 5-min averages from 60-s sampling intervals. The wind measurements are conducted more frequently to account for turbulence. All meteorological instruments are listed in figure 2.2.

The measurements are used to build a comprehensive and robust set of statistics, including means and deviations for different time periods, as well as diurnal courses for all relevant parameters. Case studies are performed by considering various types of weather conditions. Based on the visibility data, statistics of fog occurrence are built. Fog duration is calculated for each day and for different periods during a day. Fog events are classified into different fog density classes. A regression analysis is performed between fog duration and various meteorological parameters to reveal the atmospheric parameters involved in the entire fog formation process. For a detailed description of the data analyses refer to chapter 3.2.3.

The variables derived from the scatterometer measurements also feed into the fog detection algorithm in work package 3.

Work package 2. To reveal the meteorological conditions in ridge and valley locations and to address H2, parallel measurements with mobile climate stations mounted in canopy trees are conducted. The single rope technique is used to access the tree crowns (Mitchell et al. 2002).

Upper slopes and hills are expected to be free of fog and hence covered by ordinary tropical rain forest. Cold air pools likely form at night in these areas. In contrast, valley locations of putative lowland cloud forest are supposed to be frequently exposed to fog. The meteorological instruments are directly installed into trees to investigate the canopy microclimate. In order to derive the microclimatological information related to the different zonation of epiphytic plants, the instruments are placed in two height zones (inner tree crown, outer canopy). These height zones are ecologically relevant because of general differences in epiphyte diversity; see Johansson (1974) and Gradstein et al. (2003).

Air temperature, relative humidity and global radiation are measured in both height zones to analyse the climatic coupling between these zones and to elaborate the different living conditions for epiphytes. At the interface between the forest canopy and the atmosphere, wind speed and wind direction are derived by a 2D-sonic anemometer. Rainfall is measured with a rain gauge.

(36)

2 Conceptual Design

In combination with the fog measurements at COPAS, the additional vertical information of the mobile measurements in the valley helps to reveal the fog dynamics by showing to what extent the canopy is decoupled from the tree crown during times of potential fog formation. The comparison between the measurements at the hill site is important since a different canopy microclimate is expected.

At the sites where mobile meteorological measurements are undertaken, the trees are sampled for epiphyte biomass, abundance and diversity. This work is carried out by the University of Göttingen and the derived botanical inventories are used for the interpretation of the meteorological results. A principal component analysis is performed in order to elaborate the relationship among relative humidity, air temperature and epiphyte diversity.

Work package 3. Remotely-sensed imagery from satellites can provide full spatio-temporal coverage which is needed to reveal fog occurrence throughout French Guiana. Testing H1 requires the development of a satellite-based fog-detection algorithm capable of detecting fog in tropical lowland forests.

Since fog in the study area is hypothesized to occur only at night and early morning, night-time satellite data providing infrared channels have to be used. Over the past decades, specific algorithms have been developed for the detection of fog during night-time (e.g. Bendix, 2002; Eyre et al., 1984; Turner et al., 1986). The basis for most of these schemes are differences in the emissivity of small droplet water clouds in the wavelength around 3.9 µm versus the emissivity around 10.8 µm. Brightness temperature differences can be used to detect pixels of low stratiform cloudiness.

The existing algorithms have been developed for regions in the mid and high latitudes and have to be adapted for the detection of fog over tropical lowlands. In particular, the algorithm has to be adapted for fog detection at sub-pixel levels in order to account for the small-scale nature of fog in the study area. The reliable detection of small patches of fog in sub-pixel regions is generally considered to be difficult (e.g. Cermak et al. 2009). The region of French Guiana serves as a test case for a novel fog detection scheme.

To address these challenges, data from the MODIS instrument, flying on-board the polar orbiting Aqua satellite, is used. It offers a 1 km resolution for the infrared channels for wavelengths at 3.660 - 3.840 μm, 10.780 - 11.280 μm and 11.770 - 12.270 μm, which are essential for fog detection.

Four years of MODIS Level 1b calibrated radiances are used, corresponding to the period of meteorological measurements. In addition, the fog detection scheme requires the MODIS Level 2 Total Precipitation Water product and a Digital Elevation Model derived by SRTM data. Radiative transfer calculations

(37)

2 Conceptual Design (RTC) are conducted to derive thresholds functions for varying TPW amounts and sub-pixel cloudiness. A discriminant analysis is performed for the assessment of these functions. Basis for the validation of the classification results are the scatterometer measurements at the study site. The algorithm development including the preprocessing of the satellite data is presented in chapter 6.3.

The novel fog detection scheme is used to derive the first fog frequency maps of French Guiana. The analysis of fog dynamics in French Guiana is expected to answer whether lowland fog development, and hence the precondition for the formation of lowland cloud forest, is a local phenomenon or if it occurs throughout the country. figure 2.2 gives an overview on the instruments, data and methods used in this thesis.

(38)

References

References

Bendix, J., 2002. A satellite-based climatology of fog and low-level stratus in Germany and adjacent areas. Atmospheric Research 64, 3-18.

Cermak, J., Eastman, R.M., Bendix, J., Warren, S.G., 2009. European climatology of fog and low stratus based on geostationary satellite observations. Quarterly Journal of the Royal Meteorological Society 135, 2125-2130.

Charles-Dominique, P., Gottsberger, C., Freiberg, M.S., A.D., 2002. COPAS: a new permanent system to reach the forest canopy, in: Mitchell, A.W., Secoy, K., Jackson, T. (Eds.), The Global Canopy Handbook. Global Canopy Programme, Oxford, pp. 91-96.

Eyre, J.R., Brownscombe, J.L., Allam, R.J., 1984. Detection of fog at night using Advanced Very High-Resolution Radiometer (AVHRR) Imagery. Meteorological Magazine 113, 266-271.

Gradstein, S.R., Nadkarni, N.M., Kromer, T., Holz, I., Noske, N., 2003. A Protocol for Rapid and Representative Sampling of Vascular and Non-Vascular Epiphyte Diversity of Tropical Rain Forest. Selbyana 24, 105-111. Johansson, D., 1974. Ecology of vascular epiphytes in West African rain forest. Mitchell, A.W., Secoy, K., Jackson, T., 2002. The Global Canopy Handbook.

Techniques of Access and Study in the Forest Roof. Global Canopy Programme, Oxford.

Turner, J., Allam, R.J., Maine, D.R., 1986. A Case Study of the Detection of Fog at Night Using Channel 3and 4 On the Advanced Very High Resolution Radiometer (AVHRR). Meteorological Magazine 115, 285-290.

(39)

3

Canopy level fog occurrence in a

tropical lowland forest of French

Guiana as a prerequisite for high

epiphyte diversity

This chapter is published in Agricultural and Forest Meteorology, 151, 290-300 (2011).

Received: 4 February 2010 / Accepted: 31 October 2010

(40)

3 Canopy level fog occurence

Canopy level fog occurrence in a tropical

lowland forest of French Guiana as a

prerequisite for high epiphyte diversity

Obregon, A.

(1)

, Gehrig-Downie, C.

(2)

, Gradstein, S.R.

(2)

,

Rollenbeck, R.

(1)

, and J. Bendix

(1)

(1) Laboratory for Climatology and Remote Sensing, Department of Geography, University of Marburg, Germany

(2) Department of Systematic Botany, Institute of Plant Sciences, University of Goettingen, Germany

Abstract Fog frequency and the meteorological processes leading to fog formation have never been studied in depth in tropical lowland forest areas. This study provides detailed evidence of frequent fog occurrence in lowland valleys of central French Guiana. Fog frequency showed a clear diurnal course, with a maximum before sunrise; average fog duration was 4.6 h. The diurnal course of visibility was positively correlated with the diurnal course of humidity in the above-canopy air. Fog persistence correlated significantly with atmospheric parameters during the dry season, but not during the rainy season. The main trigger of fog development in the lowland forest seemed to be precipitation, leading to higher soil moisture, greater evapotranspiration and, thus, higher water content of air. An increasing temperature difference between valley and hill sites after sunset, together with more frequent down-slope winds during nights with long fog periods, points at some influence of katabatic flows. The frequent occurrence of fog in the valleys correlated with significantly higher epiphyte diversities in valley forests as compared to hill forests, and supported the occurrence of the hitherto undescribed, epiphyte-rich “tropical lowland cloud forest” (LCF) in the valleys. The higher epiphyte diversity in LCF coincided with significantly higher relative air humidity in LCF than in hill forest. The ecological benefits of fog for the epiphytes in LCF are surplus of moisture and delayed onset of the stress period, particularly in the dry season.

(41)

3.1 Introduction

3.1 Introduction

Fog as defined by a horizontal visibility of ≤1km (Roach 1994) is normally the result of very low cloud formation in the atmospheric boundary layer. It is an obstacle to traffic but, at the same time, can provide significant amounts of water to moistened surfaces by fog water interception. While a great number of studies on fog are available for the outer Tropics (e.g. Wanner & Kunz 1983, Bendix 2002, Gultepe et al. 2007, Cermak & Bendix 2008, Eugster 2008) and for tropical montane cloud forests (e.g. Hamilton & Bruijnzeel 2000, Bendix et al. 2008), investigations in tropical lowland forest are virtually lacking. Low visibilities in Midlatitudes often occur in wintertime, frequently related to radiation fog which is formed predominately in valleys and basins due to nocturnal outgoing radiation and cooling of the boundary layer air to the saturation point. In tropical mountain forests, low visibility (and thus fog) is often the result of cloud formation/advection at/to the windward slopes, mostly at altitudes >500m a.s.l. (e.g. Shuttleworth 1977, Cavelier & Goldstein 1989, Hamilton et al. 1995, Ataroff 1998, Bruijnzeel 2001, Rollenbeck et al. 2006). In the lowland tropics and subtropics, fog occurrence is well documented for coastal waters, particularly in areas of cold upwelling of the coast (e.g. Olivier 1995, Cereceda et al. 2002, Shanyengana et al. 2002, Osses et al. 2005). In contrast, knowledge of temporal and spatial fog dynamics in tropical lowland forests is poor. Observations in the Amazon forest of Brazil suggest that radiation fog might occur in lowland river valleys (Bastable et al. 1993, Klockow & Traga 1998). Large scale atmospheric circulation patterns related to fog formation in eastern Brazil were recently described by Fedorova et al. (2008). However, no comprehensive study is available on the interaction of energy balance, local breeze pattern and fog occurrence in the tropical lowland forest.

Fog is generally considered to be a major driver of the diversity of epiphytic organisms in tropical forests (e.g., Grubb & Whitmore 1966, Nadkarni 1984, 2010). It is expected that the surplus of moisture by the interception of fog water in the forest canopy shortens the duration of desiccation of the epiphytic plants (mosses, liverworts, lichens, ferns, flowering plants) dwelling in the canopy, thus enhancing their photosynthetic activity and growth.

The coincidence of fog and high epiphyte diversity is well documented for tropical mountain forests (e.g. Hamilton et al. 1995, Richards 1996) but has not been recorded in tropical lowland forest, until recently by Gradstein (2006) and Gradstein et al. (2010).

Based on findings of high epiphytic bryophyte richness of lowland forest in valleys of central French Guiana (Gradstein 2006), a new unique ecological

(42)

3 Canopy level fog occurence

habitate type, the “Tropical Lowland Cloud Forest” (LCF), is proposed, with exceptionally high richness of epiphytic bryophytes resembling epiphyte richness in tropical mountain forests. “Tropical Lowland Rain Forest” (LRF) with lower epiphyte richness is restricted to slope and hill sites. However, the specific mechanisms of low cloud formation are hitherto unknown. In this paper, we hypothesize that LCF is mainly related to radiation type valley fog formation, which is poorly documented for tropical lowland areas. To test the hypothesis, the aims of the current study are:

1. To investigate fog frequency differences in valleys and ridges. 2. To analyze meteorological processes leading to fog formation.

3. To shed light on the relation between fog occurrence and species richness of LCF epiphytic vegetation.

The results are expected to be of great importance for the hydrological cycle of the Amazon lowland forest in a broader way because a high density of epiphytic vegetation in canopies can alter canopy storage capacity and interception efficiency significantly (e.g Veneklaas et al. 1990).

3.2

Materials and methods

3.2.1

Study area

The study was performed at research station Saut Pararé, Nouragues Natural Reserve (4°2'30" N, 52°40'30" W, 75 m a.s.l.) in the valley of the Arataye river, ca. 100 km inland from Atlantic coast (figure 3.1a).

The area is covered by dense, nearly undisturbed primary lowland forest, described as “old-growth terra firme dense rainforest” (Poncy et al. 2001). The forest is tall, with canopy trees reaching heights of approximately 50 m. Natural disturbances such as tree falls lead to canopy gaps in some places (Van der Meer & Bongers 2001). The forest is dominated by Burseraceae trees and represents one of the two main French Guianan forest types described by Sabatier and Prévost (1990) differing by the dominant tree family. The terrain is undulating, with many small hills and creeks, varying in altitude from 50 to 200 m a.s.l.

In terms of meteorological measurements, the Guiana Shield is one of the poorest monitored regions in the western hemisphere (Hammond 2005). Generally, annual precipitation shows a latitudinal gradient from coastal regions to inland locations. Average annual precipitation ranges between 3500 mm at the Atlantic coast (Cayenne) and 2500 mm at Maripasoula on the border to Suriname (150 km west of the study site). The wind regime in 10 m (sigma-995

(43)

3.2 Materials and methods level) in the wider study area based on an analysis of NCEP–NCAR reanalysis data (figure 3.2, for data see Kalnay et al. 1996) clearly shows that the wind direction is dominated by the easterly trade winds throughout the whole year at all times of the day.

Figure 3.1. Measurement approach at the study site. (a) Location of the study site; (b) layout and visualization of hypothesis; (c) placement of mobile sensors in the canopy; (d)

(44)

3 Canopy level fog occurence

Meteorological measurements in the Nouragues Natural Reserve are being conducted since 1996 at the “Inselberg” field station, at a distance of 7 km from the study site (Charles-Dominique 2001). Grimaldi & Riéra (2001) provided meteorological data for November 1987–December 1996. Average annual precipitation at Inselberg field station is 2990 mm, with 310 days of rainfall. A relative dry season occurs from mid-August to mid-November, with less than 100 mm monthly rainfall during September–October, and a secondary decrease in rainfall between February and April. The rainy season lasts from November until August, with May being the wettest month (407 mm total rainfall). Air temperature exhibits little seasonal variation, mean monthly values ranging from 25.5°C in January to 27.5°C in October.

Figure 3.2. General streamflow patterns in 10m (sigma-995 level) from NCEP reanalysis data at different times of day for a) dry season (August–October 2007) and b)

rainy season (May–July 2008)

3.2.2

Experimental setup

The measurement approach is illustrated in figure 3.1. Permanent and mobile meteorological observations were conducted here between 2007 and 2009. In the valley of the Arataye river, data of important meteorological parameters were gathered directly above the canopy (45 m above ground) by installation of a climate station on one pylon of the “Canopy Operating Permanent Access System” (COPAS, 75 m a.s.l.; figure 3.1d) (Charles-Dominique et al. 2002). The station encompassed a psychrometer (H301, Vector Instruments), a 3D-sonic anemometer (CSAT3, Campbell Sci.), a net radiometer (CNR1, Kipp and Zonen) and a leaf wetness sensor (Model 237, Campbell Sci.). Horizontal visibility (and thus fog occurrence) and rainfall was observed by using a visibility and present weather sensor (HSS VPF-730, Biral). All instruments were operated by solar

Referenzen

ÄHNLICHE DOKUMENTE

To assess the influence of variability in forest and abiotic parameters on the pre- dicted level of potential ecosystem service supply per forest types, we predicted the

tentaculata distribution was assessed using sampling points in databases (Global Biodiversity Information Facility – Gbif 2016, Iucn 2016) and literature (Dunn 1942, Maciel

Therefore, the proposed strategy consists in (i) the identification and removal of cloud and cloud shadow pixels in order to increase the temporal density of analyzable images by

Under this scenario, by the end of the 21st century the most favorable conditions for forest growth will remain in only 5.7% of Ukraine, and 57.3% of the country will have a dry

The largest area of forest decline due to emissions is situated in Siberian Subarctics, in the south-western part of Taimyr peninsula around Norilsk polymetallic industrial

In a certain sense, the models which were used (and especially the uneven-aged stands model) should be termed &#34;data-orientedn models, meaning that they have

Ditri (ed.) Acid Precipitition: Effects o n Acid Precipitation, Ann Arbor, Science, Ann Arbor, Michigan... Systems Metho- dologies and

Effects of slower afforestation (the year 2000).. Optimal Possibilities for the Satisfaction of Timber Demand a t National Economic Level in Hungary. Laxen-